Retrospective clinical analysis of circulating tumor DNA (ctDNA)-based molecular response (MR) and baseline blood-based tumor mutational burden (bTMB) for monitoring response in phase 3 trial of bintrafusp alfa vs. pembrolizumab treatment of non-small cell lung cancer (NSCLC)

9098 Background: The clinical utility of ctDNA-based approaches, such as bTMB and MR, to monitor and predict response during chemo-immunotherapy was evaluated using longitudinal cohort samples from a Phase 3 study of bintrafusp alfa versus pembrolizumab as a first-line treatment in patients with PD-L1 expressing advanced NSCLC. Methods: ctDNA was collected at baseline and day 42 after treatment with bintrafusp alfa (Q2W, 1200mg) or pembrolizumab (Q3W, 200mg). The GuardantOMNI liquid biopsy (LBx) assay was used to detect somatic alterations in 497 genes and generate bTMB from baseline, and MR scores from baseline and day 42 (n = 424 samples). bTMB and somatic alterations influencing response were explored. MR scores were calculated using the validated Guardant Response algorithm. Associations between ctDNA metrics and PFS were assessed. Results: 418/424 samples passed sequencing QC; of these, somatic mutations were detected in 236/237 baseline samples and 177/181 day 42 samples. Of 212 NSCLC patients with eligible bTMB results at baseline, 104 (49.1%) patients received bintrafusp alpha and 108 (50.9%) received pembrolizumab. bTMB high (bTMB-H) is defined as >20mut/MB. In the bintrafusp alpha arm, PFS was significantly longer in patients with bTMB-H (median 8.3 months vs 2.7 months, p = 0.00086). In the pembrolizumab arm, PFS was also longer in patients with bTMB-H (median 5.65 months vs 5.5 months, p = 0.0898). Of 132 NSCLC patients with eligible MR results, 67 (50.8%) patients received bintrafusp alpha and 65 (49.2%) patients received pembrolizumab. In the bintrafusp alpha arm, molecular responders had significantly longer PFS (median 7.9 months vs 4.2 months, p = 0.00042). In the pembrolizumab arm, molecular responders also had significantly longer PFS (median 8.95 months vs 4.1 months, p = 0.00054). Conclusions: ctDNA analysis from plasma samples supported MR assessment in patients treated with ICI, indicating its utility as an adjunct to RECIST in monitoring of tumor response. Blood-based TMB-high was associated with immunotherapy treatment benefit. Analysis of b-TMB and MR allowed identification of patients with improved PFS in both treatment arms. Investigation of treatment specific TMB cut off selection and potential clinical utility of defined somatic mutations is ongoing. Clinical trial information: NCT03631706 .